You need that mathematical exposure Not very difficult but we don’t come across those mathematics in our day to day life so we find it difficult Otherwise you need to know the branches You need to know your stadium on You have one more thing They’re forgetting All the stadiums sit on each other like the sandwich which actually empower the support vector machine over Similarly in your network we have two heroes one of school a studio that the one is called Magro to I don’t know what he said on whatever he said It’s similar to what we do in your Littles On these two is born Give the mathematical power tow.
This new little becomes so far In fact newly work is also a bonus universal function approximately There’s a very strong word Okay Universum function approximately when we build models When the build models what are we doing by every building Models would have a chilling a building Mortars Yeah we’re predicting We’re estimating what’s happened Correct Right All models they represent certain processes in Dorian world The processes could be tangible or intangible processes If a building models to predict mpg off a car Given that tribute to the car and predicting the behavior of a tangent process for the car.
If I’m trying to build a bottle and see what is the stock price tomorrow in Bombay Stock Exchange that I’m trying to build a model to replicate the behavior of Bombay Stock Exchange is intangible process So if you look at it this way all mortals are nothing but representations off processes in the real world If we’re able to represent the process is immortal then by understanding that if I give this import borders I’ll perplex you to be I can predict the behavior of the process how it’s going to be here I take advantage of the beauty that knowledge right.
Now the question comes what is the process A process is a collection of vilifying steps which act on in input I’m given output given over Okay process huh The process tells you the relationship with no sex and inputs on this relationship is mathematically shown Us Why is a functional x one and x two Isn’t this what you doing Ah linear regression or any other thing any of the number Why is a function of these to this function Is what is a process on models that build They represent the processes that means on the models that were built in mathematics circle functions and when he said Neural network is a universal function Approximate er that means using neural network You can represent any process in the real world tangible and intangible It’s a very powerful statement.
You can’t do this with any other With new network You can You can build a more presentable on the stock exchange You can build a mortar to the present automotive the car We can build a model to find out your weather or predict the weather It’s a very powerful statement A neural network is a universal function Approximate which means it can represent any process in the world tangible loading that is a power that it enjoys and where does the power come from The power come from be stood here and supporting it and it also comes from another fact that happens in the neutrons The factors This is a Linnaeus emission What happens here Loneliness transformations.
This is linear transformations Vineyard functions all put off Lena function is transformed again using a non linear transmission and the left and right and a left Right now it is because of this nonlinear consummation Along with these two pills it begins a very powerful tool using which you can represent any process in this When you say you changed from a linear toe nonlinear function so are you changing the function itself Origin either data data data to represent to behave like a nonlinear Okay so I’m gonna raise this lot of things to business down the line People treat you really focus some kind of black art Black magic It is not If you understand that Link concepts nearly Look it’s so easy to learn and understand.
What’s going on inside Okay little let’s go and see the court Yeah it’s going to be the court record is shared with you I think this morning somewhere before Yeah Okay I’m going to start The first line is important and I from kind of simple choice This is a simple bite on statement There is a bunch of functional piece available under library court Random From there I’m importing one particular one called choice Okay What does Joyce just given a data it randomly picks up on record sensitive the algorithm OK Sex family for random selection of records on the given day from Mumbai I’m inputting Ari from numb pine Umpire All off you know what is numb by numerical python.
It has all the numerical functions in corded there in C So one of the functions that is included there is the dot function What does the dark function do Dot function is basically a multiplication and information together So when I won’t write bicycle m bone X one plus in tow x two plus even I want to do this and C is also having in our case w three will call it n three plus B When you want to do this you keep em one m toe on m three In one metrics you keep x one x two and be in another metrics Dart operation is multiplication of these two Makris and submission So the result that you get here this result is nothing but this he question output of this secret in is more to get here.
What you’re doing to get this result you multiplying the Mac prisons and something of the something them up That concept is called dot operation duty So when you do this dark it automatically does all these things and gives it that is it Entire neural network is metrics Algebra beast Okay so one of the first introduction to Patrick’s lt Brian you’re late will be the daughter operation Okay let’s come back to this So I’m importing dot for that I honestly don’t remember what I’m using random for Right Okay I’m importing numb pies and then here.
I’m going to demonstrate to you the behavior off Mega lo picks Neuron first make a low pitch Neuron did not have any concept of weights They give equal importance to all them perpetrators So what I’m doing here is just to make this court a general court just to make this court agenda record Ah this is what it’s called So what I’m doing here is I’m initializing both these weights with one.
So keep in mind make a low pitched in Iran did not have any bias live It had one lee inputs extra next to both of them I’m giving the weight off one right now And in between you have a step function this omission There’s no bias in my club It’s neuron So we don’t have any bias here So what you’re doing is not Look at this I’m initializing the weights with one Both of eight are equal one on This is the training data This training data if you observe carefully Which gate is this This is Andy Dick I’m giving an and Gate Day Does it So the new dawn on Dhere I’m defining my step function.
It’s a daughter Operation will give me this the step function after define the step function I’m defining here using Lambda operation all of your family to Lambda Okay Lambda is a function that is used only once After that it’s discarded So what I’m saying it’s Look at this carefully Input of the land dies X If you’re If you’re kind of getting confused yet you can replace this with some input of the Lambda Is X If the X is less than equal listeningto then go to zero All put a zero If the summation is listening to otherwise output of one Do you understand this light Are you all with me on this Okay.